We recently shared with customers our plans to deprecate Qodo’s existing code generation features. We want to be transparent about why we’re making this decision and what it means for the future of our platform.

Users will no longer be able to use the autocomplete functionality in Qodo’s IDE plugin, as well as the chat for code generation purposes. This may come as a surprise, as some customers have told us they genuinely value Qodo’s code generation capabilities, especially the quality-aware suggestions that made our approach distinct. We hear that, and we don’t take that trust lightly.

While we are deprecating code generation capabilities, the Qodo IDE plugin still remains part of the platform and we will continue to support review capabilities inside the IDE.

• Pre-PR reviews in the IDE: Qodo reviews your code as you write it, before you even open a pull request.
• Scanning committed and uncommitted changes: Qodo evaluates both staged changes and in-progress work, helping you catch issues early, before they reach the PR.
• One-click resolution workflows: From within Qodo, you can resolve issues with a single click. Qodo passes the prompt to your preferred coding agent, allowing you to fix issues directly without leaving your workflow.

From day one, Qodo was built on a clear conviction: the hard part of software development is not writing more code. It is verifying that the code works. It is understanding intent, enforcing architectural decisions, upholding standards, and ensuring quality at scale. From our first PR review and testing agents to today building the most advanced AI platform for code review and code governance, that vision has always been at the core of Qodo.

Why We’re Making This Decision

When we released a code generation product, we believed it was a natural complement to our core code review offering. Code review and code generation are a yin and yang. Code generation is about producing output. Code review, governance, and quality are about understanding intent, enforcing standards, and making sure what gets shipped actually meets the bar.

However, they are distinct problems that require different architectures, different AI harnessing, different user experiences, and different kinds of domain expertise.

At the same time, the landscape around code generation has shifted since we first introduced our IDE plugin for code generation.

Commoditization of code generation

High-fidelity code generation has rapidly become a commodity. Today, it is driven by generalized models that are extremely expensive to operate but provide increasingly similar outputs across the market. By stepping back from the generation “arms race,” we can focus our resources toward the 99% precision and recall required for code integrity.

The Need for Independent Verification

We believe that the tool generating code should not be the same tool reviewing and verifying it. This is a principle the market is increasingly confirming as organizations prioritize AI safety and trust. In any high-stakes engineering environment, whether it is finance or aerospace, you never let the builder be their own inspector. When one system handles both creation and validation, it creates a biased feedback loop where the AI is essentially “grading its own homework.”

Code generation has developed into its own demanding discipline

Code generation has become a demanding discipline in its own right. Building a strong experience today requires careful work across interaction patterns, model behavior, and IDE workflows. These are not trivial, and doing them well requires dedicated focus.

At the same time, effective code generation still depends heavily on autocomplete, iterative prompting, and close developer involvement. These patterns reflect the current state of the technology. We believe they are not the end state.

The Next Frontier

As systems for code review, governance, and quality improve, the way developers interact with code will shift. The focus will move away from managing generated code line by line, and toward defining intent, planning tasks, and setting clear standards that systems can execute and verify.

The missing piece is the quality layer that makes this possible.

That is the problem we set out to solve, and it is only becoming more important. Automated code governance, the ability to verify that code meets your standards with near-perfect precision and recall, across hundreds of quality dimensions, is still the frontier. It’s the bottleneck standing between where we are today and truly autonomous software development. And it’s where we can have the greatest impact.

What This Means for You

If you’ve been using Qodo’s code generation features, we encourage you to explore the many strong tools available in the market for code generation and AI-assisted coding. Qodo’s code review can work alongside any copilot and coding assistant. Our code review and governance platform continues to expand, and we’re confident the value it provides will only grow from here.

Our new agentic code review system recently achieved top performance on Martian’s Code Review Bench, demonstrating industry-leading recall and precision in finding bugs and issues.

Looking Ahead

We’re not saying we’re walking away from code generation forever. Once our code review and governance platform reaches the level of precision we’re building toward, we will consider re-entering the code generation space. But when we do, it will probably look very different than how it looks today.

We will be happy to hear your thoughts, suggestions or concerns about this move by Qodo. Reach out to us at [email protected]